Study On The Correction Method Of Air Conditioning Load Prediction Model At The Building Planning Stage

The basis for the energy supply of regional buildings is the accurate prediction of the air conditioning load of regional buildings.In the stage of building planning,only limited parameters are extracted from the limited amount of information.The difficulty of the research is that the design scheme is uncertain and the building information is incomplete.Generally,the area index method is often used in engineering,and this static load forecasting method will cause excessive overload prediction.Over prediction cause mismatched installed capacity of the energy station and waste of initial investment.In some research,the energy consumption simulation method is often used to calculate the air conditioning load in the planning stage.The problem is that the parameters in the building planning stage are very limited.In order to apply the energy consumption simulation software to realize the load forecasting,a lot of assumptions can be made to make the prediction accuracy greatly reduced.According to the characteristics of the building planning stage,a modified radiation time series method which can be applied to the building planning stage is proposed in this paper.The three key problems in the radiant time series method are solved,whicn are the parameters selection of heat conduction time series,radiant time series and convection radiation ratio.As we all know,the forecast result of radiant time series method is slightly large and the window-wall ratio in the planning stage is uncertain.So a correction factor,the radiation loss correction factor,is defined in this paper,Its physical meaning is to consider the radiant heat radiated to the outside through the glass.This correction can make the predicted value more accurate.It solves the problem that the original RTSM has a large load forecast due to the fact that the radiant heat is transmitted from the indoor to the outside through the outer surface of the transparent building.By example,the accuracy of the radiated RTSM for different types of buildings is improved compared with the original RTSM.Generally,the accuracy can be improved by about 2% from the prediction indicators cv-rmse and mape.In order to compare with the improved RTSM,an RC thermal network model method which is based on the standard building model and particle swarm algorithm identification was proposed.The standard building model was built to divide the building into empty buildings and full buildings while considering different climate zones.Based on the particle swarm optimization algorithm,the standard building model is identified and the effective heat capacity value is obtained.The equivalent thermal resistance is obtained by the equivalent thermal resistance theory.Finally,the third-order equivalent model of the building planning stage is established for air conditioning load prediction.In this paper,the tests were carried out in cold areas and hot summer and warm winter areas respectively,and the individual buildings and buildings were verified separately.The results show that the improved RTSM and RC thermal network model methods have good accuracy in predicting daily peak load,both for single buildings and for buildings.In the hourly load forecasting,the improved RTSM has a poor prediction effect,and the RC thermal network model method has superiority in addition to the deviation of the single building thermal load prediction effect.This indicates that RTSM is more suitable for the prediction of peak load during the building planning stage,and the RC network model method can be used in the building planning stage to effectively predict the peak load and accurately grasp the dynamic characteristics of building air conditioning load.